Filters








14 Hits in 3.8 sec

Gunrock: a high-performance graph processing library on the GPU

Yangzihao Wang, Andrew Davidson, Yuechao Pan, Yuduo Wu, Andy Riffel, John D. Owens
2015 Proceedings of the 20th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming - PPoPP 2015  
For large-scale graph analytics on the GPU, the irregularity of data access/control flow and the complexity of programming GPUs have been two significant challenges for developing a programmable high-performance  ...  We present a novel data-centric abstraction for graph operations that allows programmers to develop graph primitives at a high level of abstraction while simultaneously delivering high performance.  ...  Acknowledgments We thank Joe Mako for providing the speedup chart design.  ... 
doi:10.1145/2688500.2688538 dblp:conf/ppopp/WangDPWRO15 fatcat:utd3goviwva2ta34pi3qorbpwu

Gunrock

Yangzihao Wang, Andrew Davidson, Yuechao Pan, Yuduo Wu, Andy Riffel, John D. Owens
2016 Proceedings of the 21st ACM SIGPLAN Symposium on Principles and Practice of Parallel Programming - PPoPP '16  
For large-scale graph analytics on the GPU, the irregularity of data access and control flow, and the complexity of programming GPUs have been two significant challenges for developing a programmable high-performance  ...  "Gunrock", our graph-processing system designed specifically for the GPU, uses a high-level, bulk-synchronous, data-centric abstraction focused on operations on a vertex or edge frontier.  ...  Acknowledgments We thank Joe Mako for providing the speedup chart design.  ... 
doi:10.1145/2851141.2851145 dblp:conf/ppopp/WangDPWRO16 fatcat:72lqk4hh3ve3hdpusd43emajoi

Performance Characterization of High-Level Programming Models for GPU Graph Analytics

Yuduo Wu, Yangzihao Wang, Yuechao Pan, Carl Yang, John D. Owens
2015 2015 IEEE International Symposium on Workload Characterization  
, and thus are key focus areas for efficient largescale graph analytics on the GPU.  ...  We analyze the impact of these critical factors through three GPU graph analytic frameworks, Gunrock, MapGraph, and VertexAPI2.  ...  For more general real-world graph analytics, developers need high-level programmable frameworks to implement various types of complex graph applications on the GPU without sacrificing much performance.  ... 
doi:10.1109/iiswc.2015.13 dblp:conf/iiswc/WuWPYO15 fatcat:ueeyfurjrjcjjiyfnqkxkjaxzq

Gunrock: GPU Graph Analytics [article]

Yangzihao Wang, Yuechao Pan, Andrew Davidson, Yuduo Wu, Carl Yang, Leyuan Wang, Muhammad Osama, Chenshan Yuan, Weitang Liu, Andy T. Riffel and John D. Owens
2017 arXiv   pre-print
For large-scale graph analytics on the GPU, the irregularity of data access and control flow, and the complexity of programming GPUs, have presented two significant challenges to developing a programmable  ...  "Gunrock", our graph-processing system designed specifically for the GPU, uses a high-level, bulk-synchronous, data-centric abstraction focused on operations on a vertex or edge frontier.  ...  Thanks to the Altair and Vega-lite teams in the Interactive Data Lab at the University of Washington for graphing help. Joe Mako provided the speedup chart design.  ... 
arXiv:1701.01170v1 fatcat:kgx3yuxsrzegvkbo6x7tz5jbba

GRapid: A compilation and runtime framework for rapid prototyping of graph applications on many-core processors

Da Li, Srimat Chakradhar, Michela Becchi
2014 2014 20th IEEE International Conference on Parallel and Distributed Systems (ICPADS)  
We propose GRapid: a compilation and runtime framework that generates efficient parallel implementations of generic graph applications for multi-core CPUs, NVIDIA GPUs and Intel Xeon Phi.  ...  Many applications use graphs to represent and analyze data, but the effective deployment of graph algorithms on many-core processors is still a challenge.  ...  Mapgraph [30] , VertexAPI2 [31] and Gunrock [32] are GPU-targeting tools for graph analytics, not compiler frameworks to generate multiple GPU code variants for generic graph.  ... 
doi:10.1109/padsw.2014.7097806 dblp:conf/icpads/LiCB14 fatcat:getyhmefjfdlzbfsy4uawh46au

GraphBLAST: A High-Performance Linear Algebra-based Graph Framework on the GPU [article]

Carl Yang, Aydin Buluc, John D. Owens
2021 arXiv   pre-print
High-performance implementations of graph algorithms are challenging to implement on new parallel hardware such as GPUs because of three challenges: (1) the difficulty of coming up with graph building  ...  The design principles described in this paper have been implemented in "GraphBLAST", the first high-performance linear algebra-based graph framework on NVIDIA GPUs that is open-source.  ...  ACKNOWLEDGMENTS We thank Yuechao Pan for valuable insight into BFS optimizations. We would like to acknowledge Scott McMillan for important feedback on early drafts of the paper.  ... 
arXiv:1908.01407v5 fatcat:yiul77hmmnfffnp5snczlu6yiu

A distributed multi-GPU system for fast graph processing

Zhihao Jia, Yongkee Kwon, Galen Shipman, Pat McCormick, Mattan Erez, Alex Aiken
2017 Proceedings of the VLDB Endowment  
We present Lux, a distributed multi-GPU system that achieves fast graph processing by exploiting the aggregate memory bandwidth of multiple GPUs and taking advantage of locality in the memory hierarchy  ...  In addition, we present a performance model that quantitatively predicts the execution times and automatically selects the runtime configurations for Lux applications.  ...  A dynamic graph repartitioning mechanism achieves good load balancing among multiple GPUs, while the performance model provides insight into improving Lux's performance.  ... 
doi:10.14778/3157794.3157799 fatcat:zug24lgxfzda3ccqllnzmraqbm

Towards Benchmarking IaaS and PaaS Clouds for Graph Analytics [chapter]

Alexandru Iosup, Mihai Capotă, Tim Hegeman, Yong Guo, Wing Lung Ngai, Ana Lucia Varbanescu, Merijn Verstraaten
2015 Lecture Notes in Computer Science  
Graphalytics focuses on the dependence of performance on the input dataset, on the analytics algorithm, and on the provisioned infrastructure.  ...  graph analytics platforms.  ...  Hassan Chafi and the Oracle Research Labs, Peter Boncz and the LDBC project, and Josep Larriba-Pey and Arnau Prat Perez, whose support has made the Graphalytics benchmark possible; and to Tilmann Rabl, for  ... 
doi:10.1007/978-3-319-20233-4_11 fatcat:yvknjm45wzbx5byrkqhy3ufige

An analysis of the graph processing landscape

Miguel E. Coimbra, Alexandre P. Francisco, Luís Veiga
2021 Journal of Big Data  
For the use-case of performing global computations over a graph, it is first ingested into a graph processing system from one of many digital representations.  ...  homogeneous or heterogeneous groups of machines) and systems dedicated to high-performance computing (HPC).  ...  Gunrock couples high-performance GPU computing primitives and optimization strategies with a high-level programming model to quickly develop new graph primitives.  ... 
doi:10.1186/s40537-021-00443-9 pmid:33850687 pmcid:PMC8033100 fatcat:vnwcn2pwszhv3detnwcv6jrthu

An analysis of the graph processing landscape [article]

Miguel E. Coimbra, Alexandre P. Francisco, Luís Veiga
2021 arXiv   pre-print
The use-case of performing global computations over a graph, it is first ingested into a graph processing system from one of many digital representations.  ...  homogeneous or heterogeneous groups of machines) and systems dedicated to high-performance computing (HPC).  ...  Gunrock couples high-performance GPU computing primitives and optimization strategies with a high-level programming model to quickly develop new graph primitives.  ... 
arXiv:1911.11624v3 fatcat:t44dfa5cvfbk7exz4s2synm5z4

Benchmarking Graph Data Management and Processing Systems: A Survey [article]

Miyuru Dayarathna, Toyotaro Suzumura
2021 arXiv   pre-print
The development of scalable, representative, and widely adopted benchmarks for graph data systems have been a question for which answers has been sought for decades.  ...  We conduct an in-depth study of the existing literature on benchmarks for graph data management and processing, covering 20 different benchmarks developed during the last 15 years.  ...  They found that architecture related issues of GPUs, issues of programming models, and structure related issues of graphs as the hindrance for achieving high performance with GPU.  ... 
arXiv:2005.12873v4 fatcat:jh3367b4vjaqbgyvaccjnxqjfi

Big Data, Simulations and HPC Convergence [chapter]

Geoffrey Fox, Judy Qiu, Shantenu Jha, Saliya Ekanayake, Supun Kamburugamuve
2016 Lecture Notes in Computer Science  
We discuss convergence software built around HPC-ABDS (High Performance Computing enhanced Apache Big Data Stack) and show how one can merge Big Data and HPC (Big Simulation) concepts into a single stack  ...  of dramatic and increasing size and sophistication.  ...  We thank Dennis Gannon for comments on an early draft.  ... 
doi:10.1007/978-3-319-49748-8_1 fatcat:u4rxtm3jzbacpeg7elwhzad454

Big Graph Analytics Platforms

Da Yan, Yingyi Bu, Yuanyuan Tian, Amol Deshpande
2017 Foundations and Trends in Databases  
MapGraph MapGraph [Fu et al., 2014] targets at high-throughput graph processing with a single GPU.  ...  Comparison of the Matrix-Based Systems In a high level, the three matrix-based graph systems share commonalities, such as exposing a matrix-based interface for graph analytics and relying on a underlying  ...  ., 2015] is a semi-external memory graph engine that stores vertex state in memory and edge lists (i.e., adjacency lists) on SSDs, and its design goal is to achieve performance comparable to an in-memory  ... 
doi:10.1561/1900000056 fatcat:ucqrtzo4q5g2lpj6dmp7jv3e5m

Efficient execution of irregular programs on heterogeneous systems [article]

Rashid Kaleem
2017
] and graph analytics [Merrill et al., 2012 , Davidson et al., 2014] MapGraph [Fu et al., 2014] explains an implementation of Pregel for GPUs.  ...  Furthermore, the performance of irregular graph programs can be very dependent on the structure of the input graph: a program that performs well for power-law graphs may perform poorly for high-diameter  ...  GraphOps [Oguntebi and Olukotun, 2016] describes a collection of hardware blocks that can be used by hardware-developers to build graph analytics applications.  ... 
doi:10.15781/t27w67n4t fatcat:qlyhnhshive7bbvl5bbx4k2aem